EarnYourCert
Open the app
Exams › AI-901 › Domain 2 of 2
AI-901 · Domain 2 of 2 · 55-60% of the exam

Implement AI solutions by using Microsoft Foundry

Domain 2 of AI-901 accounts for 55-60% of the exam, the heaviest part of the syllabus. EarnYourCert has 222 original questions for this domain, all with explanations: 135 multiple choice, 35 yes/no and 52 completion or ordering items, like the real exam.

Practice domain 2 for free →No sign-up. Choose how many questions you want.

Domain 2 sample questions

8 questions from our 222-question bank for this domain. Try to answer before revealing the solution.

1. Which option shows model output progressively as it is generated, improving perceived responsiveness?

  1. Streaming responses
  2. Provisioned region
  3. Larger temperature
  4. Longer system prompt
Show answer

Correct answer: A. Streaming returns tokens as they are produced.

2. A client app wants to use an agent already created in a Foundry project. What does it typically need?

  1. A new model deployment created for every single request
  2. The admin password of the Foundry portal and the subscription ID
  3. Only the agent's color
  4. The project endpoint, credentials and the agent ID
Show answer

Correct answer: D. Clients reference the agent in the project.

3. Which technique lets a multimodal model respond directly to a user's spoken question?

  1. Sending the audio as input to a model that supports audio
  2. Sending a screenshot of the question only
  3. Using a spreadsheet to store the audio
  4. Using only text-to-speech on the client
Show answer

Correct answer: A. Audio-capable models accept spoken prompts.

4. Which Azure Speech capability lets an app speak translated output in another language during a conversation?

  1. Key phrase extraction with summarization
  2. Language detection followed by text analytics
  3. Speech recognition without translation
  4. Speech translation with text-to-speech
Show answer

Correct answer: D. Speech translation can output translated text and speech.

5. An app must support users who speak several languages. What should the speech recognizer be configured with?

  1. A PDF template for the generated transcripts
  2. The expected recognition language(s)
  3. A higher temperature so it adapts to each user
  4. A custom neural voice for every supported user
Show answer

Correct answer: B. Recognition language settings improve accuracy.

6. A generated image will be published on a website. Which practice supports transparency?

  1. Hide the origin of the image
  2. Remove all captions
  3. Disclose that the image was AI-generated
  4. Avoid any review
Show answer

Correct answer: C. Disclosure is part of transparency.

7. Which policy applies to using AI to identify people from facial images?

  1. It is unrestricted for every Azure customer by default
  2. It never needs consent when images are stored in Azure
  3. Access is restricted and needs approval
  4. It is always free
Show answer

Correct answer: C. Facial identification is a sensitive, limited-access scenario.

8. In an extraction result, in which format are the extracted field values typically returned?

  1. A bitmap image
  2. A zipped spreadsheet only
  3. A WAV file
  4. Structured JSON
Show answer

Correct answer: D. Results contain fields as structured JSON.

Practice domain 2 for free →

Other AI-901 domains

1Identify AI concepts and capabilities40-45% AI-901: full practice exam→